| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Click here to read this article as a CME activity
Department of Surgery, University of Virginia, Charlottesville, Virginia
Accepted for publication January 16, 2009.
* Address correspondence to Dr Ailawadi, PO Box 800679, Charlottesville, VA 22908-0679 (Email: gorav{at}virginia.edu).
Presented at the Fifty-fifth Annual Meeting of the Southern Thoracic Surgical Association, Austin, TX, Nov 5–8, 2008
| CARDIOTHORACIC ANESTHESIOLOGY:
The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
|
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Methods: From 1994 to 2008, 168 patients (mean age, 61 ± 14 years; male = 72, female = 96) underwent tricuspid repair (n = 156) or replacement (n = 12). Concomitant operations were performed in 87% (146 of 168). Patients with history of cirrhosis or MELD score 15 or greater (MELD = 3.8*LN [total bilirubin] + 11.2*log normal [international normalized ratio] + 9.6*log normal [creatinine] + 6.4) were compared with patients without liver disease or MELD score less than 15. Preoperative risk, intraoperative findings, and complications including operative mortality were evaluated. Statistical analyses were performed using 
2, Fisher's exact test, and area under the curve (AUC) analyses.
Results: Patients with a history of liver disease or MELD score of 15 or greater had significantly higher mortality (18.9% [7 of 37] versus 6.1% [8 of 131], p = 0.024). To further characterize the effect of MELD, patients were stratified by MELD alone. No major differences in demographics or operation were identified between groups. Mortality increased as MELD score increased, especially when MELD score of 15 or greater (p = 0.0015). A MELD score less than 10, 10 to 14.9, 15 to 19.9, and more than 20 was associated with operative mortality of 1.9%, 6.8%, 27.3%, and 30.8%, respectively. By multivariate analysis, MELD score of 15 or greater remained strongly associated with mortality (p = 0.0021). The MELD score predicted mortality (AUC = 0.78) as well as the European System for Cardiac Operative Risk Evaluation logistic risk calculator (AUC = 0.78, p = 0.96).
Conclusions: The MELD score predicts mortality in patients undergoing tricuspid valve surgery and offers a simple and effective method of risk stratification in these patients.
According to The Society of Thoracic Surgeons (STS) database, patients undergoing tricuspid valve surgery have an operative mortality of 9.8% [1]. Large centers performing tricuspid valve surgery report an operative mortality of 8% to 13.9% [2, 3]. Despite the ease of tricuspid valve repair, tricuspid disease is associated with significant comorbidities including pulmonary hypertension and liver dysfunction. Secondary liver dysfunction is thought to occur as a result of passive hepatic congestion from tricuspid disease.
Risk scoring systems have been used to compare patients' risks, to assess performance measures, and to audit quality outcomes [4]. Existing risk assessment tools in cardiac surgery, including the STS risk prediction model and the European System for Cardiac Operative Risk Evaluation (EuroSCORE), account for pulmonary hypertension but do not account for liver dysfunction. Although liver disease is a cited risk factor for mortality and complications after cardiac surgery [5, 6], there is no method to adjust for liver disease in the current risk models.
The Model for End-Stage Liver Disease (MELD) score is used to stratify patients awaiting liver transplantation [7], and its calculation is dependent on three variables: international normalized ratio (INR), total bilirubin, and creatinine. Although designed for patients with primary liver disease, MELD has been shown to predict mortality for patients with liver dysfunction undergoing nontransplant abdominal surgery [8–10]. The MELD score has been seldom utilized to assess the risk of cardiac surgery in patients with liver disease and, in particular, has never been utilized to stratify risk in patients with tricuspid valve disease. The primary objective of this study was to determine if liver disease affects outcomes with tricuspid valve surgery. Further, we hypothesized that MELD could be used to stratify patients undergoing tricuspid surgery and that MELD could predict mortality as well as EuroSCORE.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
All patients undergoing tricuspid valve operation at our institution are entered prospectively into the STS database. A retrospective review was performed of all tricuspid valve operations from January 1994 to March 2008. A total of 168 patients (mean age, 61 ± 14 years; male = 72, female = 96) underwent tricuspid valve operation.
To examine the significance of liver disease in patients undergoing tricuspid valve surgery, patients were separated based on whether or not they had significant liver disease as defined by (1) history of cirrhosis, or (2) preoperative MELD score of 15 or more.
The MELD score is defined by the equation: MELD = 3.8*LN (total bilirubin) + 11.2*LN (INR) + 9.6*LN (creatinine) + 6.4, where LN = log normal, INR = international national ratio.
Demographic, preoperative comorbidities, and operative variables were compared between the two groups. Preoperative renal failure was defined by creatinine greater than 2.0 mg/dL according to the STS database. The INR utilized for the MELD analysis was drawn on the morning of surgery. A total of 74 patients (44%) were on coumadin preoperatively, most often for atrial fibrillation. Elective patients had their coumadin held at least 5 days preoperatively, allowing for recovery of their INR. Urgent and emergent cases had reversal of their coumadin with fresh frozen plasma. Outcomes measured included operative mortality and perioperative complications.
To further characterize the effect of preoperative MELD in patients undergoing tricuspid valve surgery, we evaluated the effect of preoperative MELD on outcome. Because testing preoperative liver function tests (total bilirubin) is not routinely collected with STS data, preoperative MELD was able to be calculated in 72.6% (122 of 168) of all patients. Patients were stratified by their preoperative MELD score into four groups: less than 10, 10 to 14.9, 15 to 19.9, and 20 or more. Preoperative variables, intraoperative findings, and complications including operative mortality were evaluated. Univariate and multivariate analyses were performed to identify risk factors related to mortality.
Finally, preoperative MELD was assessed as a risk stratification tool. A logistic EuroSCORE was calculated on all patients and was compared with MELD score. The STS model does not calculate a predicted risk for patients undergoing tricuspid valve surgery. Therefore, it could not be used for comparison to MELD or EuroSCORE in this group of patients.
Statistical Analysis
Patient demographics, preoperative factors, intraoperative factors, and outcomes were first compared using univariate techniques. Continuous variables were compared using the Wilcoxon two-sample test for cases where only two groups were being compared, higher order comparisons were performed with the Kruskal-Wallis test. Categorical variables were compared with the 2 test or Fisher's exact test, where appropriate. The capabilities of MELD to predict 30-day mortality were compared with the EuroSCORE logistic model by calculating the receiver operating characteristic area under the curve (AUC) for each and comparing them using techniques described by DeLong and colleagues [11]. This analysis was performed on all patients with a preoperative MELD score.
Lastly, a multivariate logistic regression analysis was performed examining the outcome of operative (30-day) mortality. Variables entered into the model included those risk factors with significant or near significant (p < 0.1) influence on mortality by univariate analysis. Data manipulation and analysis were performed using SAS 9.1.3 (SAS Institue, Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Effect of Liver Disease on Tricuspid Valve Surgery
Patients with significant liver disease (defined by a history of cirrhosis or MELD score of 15 or greater, or both) comprised 22.0% (n = 37) of patients undergoing tricuspid surgery. Of these, 19 had MELD score of 15 or greater and the remainder were included based on history of cirrhosis. The etiology of liver disease for these patients included 13 patients (35%) with alcoholic liver disease, 4 patients (11%) with hepatitis B, and 5 patients (14%) with hepatitis C (Table 1). Compared with patients without significant liver disease, patients with liver disease more often were younger, male, and had a diagnosis of endocarditis. Patients with significant liver disease were more likely to have preoperative renal failure (51.4% versus 13.0%, p < 0.0001). Other preoperative risk factors were no different including age, ejection fraction, congestive heart failure.
|
Patients with liver disease had worse outcomes. Postoperative sepsis, gastrointestinal complications, and multisystem organ failure were significantly more common in patients with liver disease. The incidence of patients with new-onset renal failure requiring dialysis was also higher among patients with liver disease. Importantly, patients with liver disease had worse a mortality rate than did patients without liver disease (18.9% versus 6.1%, p = 0.024).
Effect of MELD on Tricuspid Valve Surgery
Patients with preoperative MELD were stratified by MELD and compared (122 of 168 patients). Preoperative and intraoperative variables are listed in Table 2. Patients with higher MELD were more likely to have endocarditis (p = 0.021) and peripheral vascular disease (p < 0.0005). Other preoperative variables and concomitant procedures were similar across the groups. Cardiopulmonary bypass time increased as MELD increased.
|
|
|
Comparison of MELD to EuroSCORE
A comparison of the receiver operating characteristic plots for the MELD score (AUC = 0.78) and the EuroSCORE logistic model (AUC = 0.78) were similar (p = 0.96; Fig 2).
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Hepatic dysfunction is known risk factor for surgical candidates. Various risk assessment tools have been developed to stratify the degree of liver disease. The Childs-Turcotte-Pugh classification (CTP) was developed to assess the operative risk in patients undergoing portosystemic shunts [12, 13]. The CTP classification model subsequently became the leading assessment method of classify functional hepatic status. For surgical patients, the CTP model was extrapolated to predict postoperative outcomes in other patients with cirrhosis [14]. One inherent problem with the CTP classification system, however, was the inclusion of largely subjective parameters (such as presence of ascites, nutritional status, and degree of encephalopathy). Consequently, in 2000, a new risk assessment method was developed for patients with liver disease. The Mayo End-Stage Liver Disease score was originally developed to assess the functional liver status in patients undergoing transjugular intrahepatic portosystemic shunts [15]. Subsequently, the scoring system was renamed to the Model for End-Stage Liver Disease (MELD) when it was adopted by the United Network for Organ Sharing and utilized in the allocation of cadaveric livers [7, 16].
The MELD score has become the preferred risk assessment tool for cirrhosis in general surgical patients [8, 10, 17, 18]. Several groups have compared the CTP classification and MELD score in patients undergoing general surgical procedures [8–10]. A recent study concluded that a MELD score higher than 14 was a worse prognosticator than CTP class C for perioperative complications for intra-abdominal surgery [8]. Northup and colleagues [10] furthered these findings in 140 nontransplant surgical patients and demonstrated that, for the first time, MELD was predictive for operative mortality. In that study, mortality increased 1% to 2% for every MELD point. The CTP is subjective, whereas MELD is able to be calculated on three common blood tests. Moreover, MELD includes the effect of renal dysfunction, which is a significant comorbidity in patients with liver disease.
Liver disease is an important risk factor in patients undergoing cardiac surgery [5, 19, 20]. An association between tricuspid valve dysfunction and liver disease due to hepatic congestion has been well established [21, 22]. Suman and coworkers [18] demonstrated a strong correlation between CTP, MELD, and mortality (AUC = 0.84 and 0.87, respectively) in 44 patients primarily undergoing coronary artery bypass graft surgery. In this study, CTP score greater than 7 and MELD score greater than 13 were highly specific for mortality (92% and 89%, respectively). In contrast, Filsoufi and colleagues [20] concluded that CTP was superior to MELD score in predicting hospital mortality (p = 0.02 versus p = 0.065) in 27 cirrhotic patients undergoing cardiac surgery. Specifically, morality rates were 10%, 18%, and 67% for CTP classes A, B, and C, respectively. The results of these two studies provide conflicting results with respect to the predictive capacity of MELD score in cardiac surgery patients. Importantly, both studies analyzed small patient groups and did not have sufficient power to evaluate patients according to MELD score distribution.
Many patients with high MELD score had significant renal dysfunction. Preoperative renal disease is a well-documented risk factor in cardiac surgery and specifically in patients with tricuspid valve disease [23–25]. All existing risk calculators account for renal disease, although often as a categorical variable. Renal disease is a continuous variable affecting MELD score.
Several risk stratification schemes have been developed to predict mortality after cardiac surgery including STS Score, EuroSCORE, Parsonnet score, and the Cleveland Clinic model [26–31]. The EuroSCORE is a complex, risk stratification scheme that utilizes 17 patient-related, cardiac-related, and operation-related factors to calculate a predicted mortality rate and has gained popularity among European and Canadian surgeons [28]. Geissler and colleagues [32] compared six scoring systems used for risk stratification in 504 cardiac surgery patients and found that the EuroSCORE yielded the highest predictive value for mortality. Other recent comparisons of these models have supported the conclusion that EuroSCORE functions as a superior predictive model for risk stratification with nearly identical observed and expected mortality [32–34]. Although this scoring system appears to successfully predict postoperative mortality for the majority of cardiac surgery patients, it does not adjust for liver disease as a preoperative risk factor. Owing to the inherent multiorgan system effects of both hepatic dysfunction and cardiac surgery, a reliable predictive model of postoperative outcomes for patients with known liver diseases remains desirable. Furthermore, the EuroSCORE has not gained popularity in the United States because of the multiple variables required, including pulmonary artery pressure, which may necessitate right heart catheterization. The MELD score offers a simple method, requiring three common blood tests to calculate.
There are several limitations to this study. First, it is a retrospective study and has inherent biases. The total number of patients undergoing tricuspid surgery is admittedly small. A major limitation is in our definition of liver disease. Because liver function is not tested routinely, we defined liver disease as a history of cirrhosis or evidence of cirrhosis (by MELD). Not every patient had a preoperative MELD score, thus limiting our analysis of the effect of MELD on outcome. Importantly, patients who did not have preoperative bilirubin testing (or preoperative MELD) had equivalent mortality as patients with preoperative MELD, suggesting that there was no selection bias in those patients who had a preoperative MELD score. The time of this study is long, and during the study period there have been advances in anesthetic management related to patients with liver dysfunction. Because our operative database was used to identify patients, no medical control is available to identify the natural history of patients with tricuspid disease and liver dysfunction. Finally, it is difficult to separate the effect of renal failure from MELD, as MELD is directly dependent on renal function. It is possible renal dysfunction contributes significantly to the morbidity in tricuspid surgery.
At present, it is unknown if therapy aimed to lower preoperative MELD will have an impact on mortality. Early use of continuous hemofiltration or liver replacement therapy may alter the MELD score, but the effect on mortality for patients undergoing surgery is unclear. It is likely that patients with tricuspid valve dysfunction require correction of the structural problem in the way of valve surgery to treat the liver dysfunction.
In conclusion, in the present study, liver dysfunction was associated with poor outcome for patients undergoing tricuspid valve surgery. The MELD score predicts mortality among these patients, with especially high risk for patients with a MELD score of 15 or greater. It is likely that passive hepatic congestion lending to liver dysfunction is a significant cause of mortality after tricuspid valve surgery. Compared with EuroSCORE, the MELD score offers a simple and effective method of risk stratification in patients with tricuspid disease. Further validation of MELD as a predictive tool should be performed.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
As I understand it, the MELD score has been used extensively to predict operative mortality in patients with liver disease undergoing liver transplantation and abdominal surgery. In your manuscript you seem to use the term "liver disease" and "liver dysfunction" interchangeably. Is liver dysfunction when caused by passive congestion from heart failure, and which may be temporary, equivalent to liver dysfunction secondary to end-stage cirrhosis from other causes, which is usually permanent? Both may produce changes in creatinine, bilirubin, and INR, the components of the MELD score. Is the predictive value of the MELD score equally good in these two groups of patients? Second, one component of the MELD score is INR, and I am sure that some of your patients, as ours, must have been on Coumadin for atrial fibrillation or other reasons. We found it impossible to retrospectively identify which patients were which. Could this increased INR secondary to Coumadin administration, which in turn would have impacted the MELD score, have influenced your data and conclusions, and if so, how? Finally, has this ability to accurately predict operative mortality in these patients influenced your patient selection in any way?
I think it is apparent from your data and confirmed by ours that the MELD score was quite useful in predicting operative risk in this chronically and seriously ill group of patients. I enjoyed your presentation very much and the opportunity to review your manuscript and to discuss it here. Thank you.
DR AILAWADI: Thank you, Dr Crawford, for those thoughtful comments, and I applaud you for looking at your data so quickly, as it took us quite a bit longer to do this. The first question related to liver dysfunction versus preexisting cirrhosis. I think that is a great question, and my suspicion is that temporary liver dysfunction may not have the same effect as long-standing cirrhosis in some of these patients. However, our data set, it is just too small to really answer that question. But that is something as we bring this out to the forefront we could certainly look into further and larger series.
With respect to INR and patients on Coumadin for atrial fibrillation, all of our patients undergo routine INR testing the morning of surgery, and if they were admitted preoperatively, they had their INR corrected. So presumably in most patients it is going to be as close to their baseline as possible. That is a confounding factor that we will not be able to resolve in a retrospective study.
The MELD score has not affected our patient selection as yet, however, it is now being used in terms of discussing with patients their true risk, as liver dysfunction was never looked at previously. We do calculate MELD routinely now on patients undergoing both tricuspid valve surgery and other cardiac procedures if they have liver dysfunction.
DR STEVEN F. BOLLING (Ann Arbor, MI): Gorav, I have two questions. As you know, the vast majority of patients with TR do not come to us for surgery; we get a very small minority of them. Do you have the nonoperated mortality with MELD scores for patients with TR, in other words, what is the natural history of someone with a lot of tricuspid regurgitation and a high MELD score who we don't operate on? And do you turn patients down based on the MELD score?
For example, even with a high MELD score of 15 or 20, where the operative mortality may be 20% or 30%, what is the nonoperative mortality for that very same patient with a lot of tricuspid regurgitation and a high MELD score? Maybe it is a 100% without an operation and maybe we should consider operating on those patients. Thank you.
DR AILAWADI: Thank you, Dr Bolling. Those are great questions. As we used the STS database for this study, we do not have data on patients that did not undergo cardiac surgical procedures. So it is hard to really answer the question of medically treated patients. At this point, we are not turning down patients, but we will factor in MELD in discussing their risk.
I do want to mention that 80% of our patients had concomitant procedures. So their indication for operation was often not the tricuspid valve procedure.
DR PATRICK EUGENE PARRINO (New Orleans, LA): I was wondering if your data allowed you to demonstrate a correlation between the MELD score and the degree of right ventricular dysfunction. Was there a correlation, and if so, how many of these deaths were related more to their hepatic insufficiency versus right ventricular failure and the need for high-dose pharmacologic support postoperatively? If some of the deaths were related more to right ventricular dysfunction and pharmacologic support, did you look to see if strategies such as pump-assisted but nonarrested revascularization or beating heart tricuspid repairs may improve outcomes in the patients with high MELD scores?
DR AILAWADI: More than half of our patients had significant right ventricular dysfunction and likely played a large part in their demise. As you know, in patients with a significant right ventricular dysfunction postoperatively, it is difficult to decipher whether that was an inciting event or a result of them being so sick, and often by that time they have multisystem organ failure. We did not look at specifically whether the clamp was on or off during the time of tricuspid valve surgery—as I mentioned, 80% did have concomitant procedures—and this technique is based on the surgeon's preference.
DR VINOD H. THOURANI (Atlanta, GA): Gorav, excellent presentation. Have you been able to correlate Child's classifications with the MELD scores? We commonly use that instead of MELD for our patient selection in those with a right-sided valvular dysfunction, especially for looking at liver disease.
DR AILAWADI: That is a great question, and part of the reason why MELD has kind of taken over in the general surgical literature is because the Child-Turcotte-Pugh classification is very subjective, in terms of ascites and encephalopathy, and it does not make for a very objective method of assessing patients. For these reasons, MELD has really been touted as the preferred method to stratify patients in the general surgical literature. That is why we chose to use MELD and not Child-Turcotte-Pugh in this study.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
